| { | |
| "best_metric": 0.9976505766766339, | |
| "best_model_checkpoint": "./beit-base/checkpoint-1755", | |
| "epoch": 2.9995728321230244, | |
| "global_step": 1755, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.840909090909091e-06, | |
| "loss": 1.0185, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.681818181818182e-06, | |
| "loss": 0.8387, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 8.522727272727273e-06, | |
| "loss": 0.6429, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.1363636363636365e-05, | |
| "loss": 0.3847, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.4204545454545456e-05, | |
| "loss": 0.206, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.7045454545454546e-05, | |
| "loss": 0.1805, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9886363636363638e-05, | |
| "loss": 0.1222, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.272727272727273e-05, | |
| "loss": 0.0944, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.5568181818181817e-05, | |
| "loss": 0.076, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.8409090909090912e-05, | |
| "loss": 0.1085, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.125e-05, | |
| "loss": 0.1311, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.409090909090909e-05, | |
| "loss": 0.0749, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.6931818181818184e-05, | |
| "loss": 0.0745, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.9772727272727275e-05, | |
| "loss": 0.0682, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.261363636363637e-05, | |
| "loss": 0.0558, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.545454545454546e-05, | |
| "loss": 0.0471, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.829545454545455e-05, | |
| "loss": 0.0662, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.987333755541482e-05, | |
| "loss": 0.1774, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.955668144395187e-05, | |
| "loss": 0.1695, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.9240025332488924e-05, | |
| "loss": 0.0893, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.8923369221025965e-05, | |
| "loss": 0.096, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.860671310956302e-05, | |
| "loss": 0.0852, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.829005699810007e-05, | |
| "loss": 0.0878, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.7973400886637115e-05, | |
| "loss": 0.1237, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.765674477517416e-05, | |
| "loss": 0.0915, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.734008866371121e-05, | |
| "loss": 0.0439, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.702343255224826e-05, | |
| "loss": 0.1507, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.670677644078531e-05, | |
| "loss": 0.1061, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.6390120329322354e-05, | |
| "loss": 0.0967, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.607346421785941e-05, | |
| "loss": 0.0994, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.5756808106396456e-05, | |
| "loss": 0.0976, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.5440151994933504e-05, | |
| "loss": 0.1442, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.512349588347055e-05, | |
| "loss": 0.0735, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.48068397720076e-05, | |
| "loss": 0.0693, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.449018366054465e-05, | |
| "loss": 0.0748, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.41735275490817e-05, | |
| "loss": 0.0585, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.385687143761874e-05, | |
| "loss": 0.0963, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.35402153261558e-05, | |
| "loss": 0.0526, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.3223559214692845e-05, | |
| "loss": 0.0852, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.290690310322989e-05, | |
| "loss": 0.0973, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.259024699176694e-05, | |
| "loss": 0.0818, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.2273590880303995e-05, | |
| "loss": 0.1427, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.1956934768841036e-05, | |
| "loss": 0.1151, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.164027865737809e-05, | |
| "loss": 0.0536, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.132362254591514e-05, | |
| "loss": 0.0959, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.1006966434452187e-05, | |
| "loss": 0.0641, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.0690310322989234e-05, | |
| "loss": 0.0931, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.037365421152628e-05, | |
| "loss": 0.0418, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.005699810006333e-05, | |
| "loss": 0.0581, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.9740341988600384e-05, | |
| "loss": 0.0573, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.9423685877137425e-05, | |
| "loss": 0.0911, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.910702976567448e-05, | |
| "loss": 0.1501, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.879037365421153e-05, | |
| "loss": 0.0979, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.8473717542748576e-05, | |
| "loss": 0.0737, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.8157061431285623e-05, | |
| "loss": 0.0443, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.784040531982268e-05, | |
| "loss": 0.0838, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.752374920835972e-05, | |
| "loss": 0.0664, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.7207093096896774e-05, | |
| "loss": 0.0303, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9942332336608287, | |
| "eval_loss": 0.01858363300561905, | |
| "eval_runtime": 148.0695, | |
| "eval_samples_per_second": 31.62, | |
| "eval_steps_per_second": 3.958, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.689043698543382e-05, | |
| "loss": 0.0769, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.657378087397087e-05, | |
| "loss": 0.0621, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.625712476250792e-05, | |
| "loss": 0.0723, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.5940468651044965e-05, | |
| "loss": 0.0851, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.562381253958201e-05, | |
| "loss": 0.0889, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.530715642811907e-05, | |
| "loss": 0.0799, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.499050031665611e-05, | |
| "loss": 0.0558, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.467384420519316e-05, | |
| "loss": 0.0548, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.435718809373021e-05, | |
| "loss": 0.0844, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.404053198226726e-05, | |
| "loss": 0.0374, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.3723875870804306e-05, | |
| "loss": 0.0829, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.340721975934136e-05, | |
| "loss": 0.0581, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.30905636478784e-05, | |
| "loss": 0.073, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.2773907536415456e-05, | |
| "loss": 0.0713, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.2457251424952504e-05, | |
| "loss": 0.0627, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.214059531348955e-05, | |
| "loss": 0.1106, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.18239392020266e-05, | |
| "loss": 0.0838, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.150728309056365e-05, | |
| "loss": 0.0921, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.1190626979100695e-05, | |
| "loss": 0.0262, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.087397086763775e-05, | |
| "loss": 0.0545, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.055731475617479e-05, | |
| "loss": 0.0867, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.0240658644711845e-05, | |
| "loss": 0.091, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.992400253324889e-05, | |
| "loss": 0.0702, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.960734642178594e-05, | |
| "loss": 0.0888, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.929069031032299e-05, | |
| "loss": 0.0514, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.897403419886004e-05, | |
| "loss": 0.0762, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.8657378087397084e-05, | |
| "loss": 0.0192, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.834072197593414e-05, | |
| "loss": 0.0337, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.8024065864471183e-05, | |
| "loss": 0.0634, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.7707409753008234e-05, | |
| "loss": 0.0435, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.7390753641545285e-05, | |
| "loss": 0.0685, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.7074097530082333e-05, | |
| "loss": 0.1016, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.6757441418619384e-05, | |
| "loss": 0.0797, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.644078530715643e-05, | |
| "loss": 0.0946, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.612412919569348e-05, | |
| "loss": 0.0317, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.5807473084230528e-05, | |
| "loss": 0.0673, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.549081697276758e-05, | |
| "loss": 0.0287, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.5174160861304623e-05, | |
| "loss": 0.0273, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.4857504749841674e-05, | |
| "loss": 0.0404, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.4540848638378722e-05, | |
| "loss": 0.0524, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.422419252691577e-05, | |
| "loss": 0.0372, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.3907536415452818e-05, | |
| "loss": 0.0706, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.359088030398987e-05, | |
| "loss": 0.0451, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.3274224192526917e-05, | |
| "loss": 0.0583, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.2957568081063964e-05, | |
| "loss": 0.0233, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.2640911969601016e-05, | |
| "loss": 0.0509, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.2324255858138063e-05, | |
| "loss": 0.0619, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.200759974667511e-05, | |
| "loss": 0.0945, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.169094363521216e-05, | |
| "loss": 0.0151, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.137428752374921e-05, | |
| "loss": 0.0675, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.1057631412286258e-05, | |
| "loss": 0.0548, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.0740975300823306e-05, | |
| "loss": 0.0327, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.0424319189360357e-05, | |
| "loss": 0.1223, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.0107663077897405e-05, | |
| "loss": 0.1124, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.9791006966434452e-05, | |
| "loss": 0.0497, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.94743508549715e-05, | |
| "loss": 0.0703, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.915769474350855e-05, | |
| "loss": 0.0212, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.88410386320456e-05, | |
| "loss": 0.1229, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.8524382520582647e-05, | |
| "loss": 0.0374, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9955147372917557, | |
| "eval_loss": 0.015015287324786186, | |
| "eval_runtime": 154.7173, | |
| "eval_samples_per_second": 30.262, | |
| "eval_steps_per_second": 3.788, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.8207726409119695e-05, | |
| "loss": 0.0663, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.7891070297656746e-05, | |
| "loss": 0.0235, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7574414186193794e-05, | |
| "loss": 0.0491, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.725775807473084e-05, | |
| "loss": 0.0361, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.6941101963267893e-05, | |
| "loss": 0.0354, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.662444585180494e-05, | |
| "loss": 0.0222, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.6307789740341988e-05, | |
| "loss": 0.0769, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.5991133628879036e-05, | |
| "loss": 0.0423, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.5674477517416087e-05, | |
| "loss": 0.0301, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.5357821405953135e-05, | |
| "loss": 0.0092, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.5041165294490185e-05, | |
| "loss": 0.0443, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.4724509183027232e-05, | |
| "loss": 0.0504, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.4407853071564282e-05, | |
| "loss": 0.0229, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.409119696010133e-05, | |
| "loss": 0.077, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.3774540848638379e-05, | |
| "loss": 0.0168, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.3457884737175427e-05, | |
| "loss": 0.0509, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.3141228625712476e-05, | |
| "loss": 0.1056, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.2824572514249524e-05, | |
| "loss": 0.0267, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.2507916402786574e-05, | |
| "loss": 0.0302, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.2191260291323623e-05, | |
| "loss": 0.067, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.1874604179860673e-05, | |
| "loss": 0.0344, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.1557948068397722e-05, | |
| "loss": 0.0408, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.124129195693477e-05, | |
| "loss": 0.0313, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 1.092463584547182e-05, | |
| "loss": 0.0065, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 1.0607979734008867e-05, | |
| "loss": 0.0507, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.0291323622545917e-05, | |
| "loss": 0.0182, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.974667511082964e-06, | |
| "loss": 0.0413, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.658011399620014e-06, | |
| "loss": 0.0626, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 9.341355288157062e-06, | |
| "loss": 0.0222, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 9.024699176694111e-06, | |
| "loss": 0.0529, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 8.70804306523116e-06, | |
| "loss": 0.0042, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 8.391386953768208e-06, | |
| "loss": 0.051, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 8.074730842305258e-06, | |
| "loss": 0.0377, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.758074730842306e-06, | |
| "loss": 0.0639, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 7.441418619379355e-06, | |
| "loss": 0.0532, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 7.124762507916404e-06, | |
| "loss": 0.0601, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.808106396453452e-06, | |
| "loss": 0.0118, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.491450284990501e-06, | |
| "loss": 0.069, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 6.1747941735275496e-06, | |
| "loss": 0.1187, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.858138062064598e-06, | |
| "loss": 0.0612, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.541481950601647e-06, | |
| "loss": 0.0387, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 5.2248258391386955e-06, | |
| "loss": 0.0172, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.908169727675745e-06, | |
| "loss": 0.0591, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.5915136162127936e-06, | |
| "loss": 0.0594, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.274857504749842e-06, | |
| "loss": 0.0249, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.958201393286891e-06, | |
| "loss": 0.0707, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.6415452818239395e-06, | |
| "loss": 0.0472, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.324889170360988e-06, | |
| "loss": 0.0371, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.008233058898037e-06, | |
| "loss": 0.0076, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.6915769474350858e-06, | |
| "loss": 0.0263, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.3749208359721344e-06, | |
| "loss": 0.0256, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.058264724509183e-06, | |
| "loss": 0.0349, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.7416086130462319e-06, | |
| "loss": 0.0315, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.4249525015832807e-06, | |
| "loss": 0.0217, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.1082963901203294e-06, | |
| "loss": 0.0409, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.916402786573782e-07, | |
| "loss": 0.0194, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.749841671944269e-07, | |
| "loss": 0.0258, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.5832805573147563e-07, | |
| "loss": 0.0559, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9976505766766339, | |
| "eval_loss": 0.011637404561042786, | |
| "eval_runtime": 150.0146, | |
| "eval_samples_per_second": 31.21, | |
| "eval_steps_per_second": 3.906, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1755, | |
| "total_flos": 4.3513294045912105e+18, | |
| "train_loss": 0.08035437971183718, | |
| "train_runtime": 5091.7575, | |
| "train_samples_per_second": 11.034, | |
| "train_steps_per_second": 0.345 | |
| } | |
| ], | |
| "max_steps": 1755, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.3513294045912105e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |